Publication | International Conference on Machine Learning 2014
Wasserstein propagation for semi-supervised learning
Abstract
Wasserstein propagation for semi-supervised learning
J. Solomon, R. Rustamov, L. Guibas, Adrian Butscher
International Conference on Machine Learning 2014
Probability distributions and histograms are natural representations for product ratings, traffic measurements, and other data considered in many machine learning applications. Thus, this paper introduces a technique for graph-based semi-supervised learning of histograms, derived from the theory of optimal transportation. Our method has several properties making it suitable for this application; in particular, its behavior can be characterized by the moments and shapes of the histograms at the labeled nodes. In addition, it can be used for histograms on non-standard domains like circles, revealing a strategy for manifold-valued semi-supervised learning. We also extend this technique to related problems such as smoothing distributions on graph nodes.
Download publicationAssociated Autodesk Researchers
Justin Solomon
Stanford University
Raif Rustamov
Stanford University
Leonidas Guibas
Stanford University
Related Resources
2010
210 King Street: A Dataset for Integrated Performance AssessmentThis paper presents a Building Information Modeling (BIM) re-creation…
1998
HMDs, Caves & Chameleon: A human-centric analysis of interaction in virtual spaceThere are a various approaches to implementing virtual reality (VR)…
2022
”I don’t want to feel like I’m working in a 1960s factory”: The Practitioner Perspective on Creativity Support Tool AdoptionCreative practitioners reflect on their values to derive a value…
2009
Effective Analogical Transfer Using Biological Descriptions Retrieved with Functional and Biologically Meaningful KeywordsWhile biology is well recognized as a good source of analogies for…
Get in touch
Something pique your interest? Get in touch if you’d like to learn more about Autodesk Research, our projects, people, and potential collaboration opportunities.
Contact us